8 Replies Latest reply on Jun 11, 2013 3:08 PM by Jared Conway


    Drew Fullerton

      I ran a simple exercise where I placed a 50mmx50mm part on top of and in the center of a 100mmx100mm part (t=25mm for each part).

      I then set up 2 studies:

      1. two faces were bonded;
      2. the edges (qty=4 edges) were bonded to one face of the other part.


      Then I applied a load on one bonded edge (not a face) (the load was a point load SW didn't allow me to place a uniformaly distributed load or pressure on an edge) and discovered that:

      1. the VM stresses within the assembly were the same; and
      2. the stresses parallel to the direction of the load, in the assembly, were the same (say the load is in "Y", then then Y-normal stress/direction is parallel to the load).


      Could anyone explain why this is the case? Definitely not what I expected, here's why:


      First,  I imagine that the face to face bond contact condition could represent glue - currently within in SW I don't think it does. I state this because when I ask SW to show the vector plot of the stress, it shows up on one edge only, not on the entire area - not what I would expect.

      Further, the values of the normal stresses, according to SW vary depending on location - this I expect due to the nature of FEA and the solvers being employed; however, nowhere do I see 200,000N/(50mmx50mm) = 80MPa, the value that we are most likely to engineer the shear bond to resist


      Second, if I can identify edges that can be bonded to a surface, then I would expect them to act as welds within SW and the results be reported appropriately so that I can then size the weld based on the appropriate throat size or have SW do it for me - QED - but not at all what I am seeing.


      In summary, I would expect to see a difference between the results one sees for face vs edge bonding + I see them as representing 2 different real world applications


      Comments, thoughts and feedback most welcome

          Bert-Jan Knol

          My first thought is that the global contact set is active and set to bonded. In this case despite of bonding the four edge, the contacting faces would still be bonded, hence the same results. If not you should indeed see different results.

            Bill McEachern

            Irrespective of whether the problem was set up correctly or not the essence is that FEA (structural analysis) is a stiffness based method. Stiffness is strongly related to the "depth" of a connection. Thus, in reality, for small spans there will not be much difference between face bonded and edge bonded in the big scheme of things. When the spans get big, then the differences will emerge and the use of plug welds to reduce the effective spans is a sensible way to deal with it. Run some experiments to see what happens.

              Drew Fullerton

              Bert - too right!!! There should be an inicator for newbies such as myself that I left global bonding on!!!!

              So, with contact sets now ok; the results are indeed different.


              Having said that, I would expect that the stress on a line (an edge) would be greater than that for a bonded area. SW appears to be saying different. Using the probe tool set to 'selected entities' function and measuring normal stress only, the RMS and average values for the bonded face experiment are approx double (945 & 923) the bonded edge experiment (418 & 417).


              Any ideas why this might be?


              Bill, I appreciate that at the macro level the 'rules of engagement' that we typically use are a massive averaging that in the long run represent our real world observatins adequately. As you are aware, we use a stiffness approach when balancing moment frames; within the context of SW FEA I'm not seeing what I might expect and am trying to put what SW is telling me in context.


              Just trying to understand what SW is doing.


              Further, the SW bonded face experiment is being represented on my screen as a line - SW is showing the stresses on the loaded edge only, nowhere else - I would expect to see a stress distribution on the bonded face - further software development perhaps? Thoughts?

                • Re: FACE BONDING VS EDGE BONDING
                  Jared Conway

                  can you post a picture of your model. i can't tell if you're trying to bond the faces taht are perpendicular to the main face or if you've created splitlines that are parallel to it like the edges would be.


                  what is your end goal?


                  also, regarding the stresses, are they affected by improvements in mesh? and, what are you seeing from a displacement perspective? are they similar for the 2 cases?

                    • Re: FACE BONDING VS EDGE BONDING
                      Drew Fullerton

                      Is there a way to upload the file to this forum?

                        • Re: FACE BONDING VS EDGE BONDING
                          Jared Conway

                          when you reply, at the top right, use the advanced editor. that will let you upload a file.

                            • Re: FACE BONDING VS EDGE BONDING
                              Drew Fullerton

                              Jared, my goal is to assure myself that I know how to interpret SW sim results.

                              I am attempting to achieve this goal by setting up simple situations where I expect a certain result - if the results do not meet my preconcieved notion, then I have to understand what has happened.

                              In other words, I wish to master this beast on a small scale before I progress further.


                              In this small case, I wanted to compare a full contact bond with edge bonds, in my mind the real world equivalent of a full surface bond is glue and I would expect to see a stress dist across the entire contact surface, SW is not showing me that - I would like to understand why.


                              Whereas with edge bonds only, I would expect to see stresses on each of the lines, which I do, so the representation in SW is ok, AND I think the real world equivalent could be a fillet weld, however, I don't see any mention of edge bonds being the equivalent of real world welds in any documentation, so what are the edge bonds really?

                                • Re: FACE BONDING VS EDGE BONDING
                                  Jared Conway

                                  hi drew, you posted the assembly but not the part files. if you use the pack and go to zip option it will grab everything.


                                  can you post a picture of the stress you're seeing with the bonded face contact? i don't fully understand what you're expecting but with a bonded contact between 2 faces they are going to move together. the bond is perfect and unbreakable. so i don't know if i'd expect stress, but the deformation of the 2 plates bonded togheter should be less than if you applied the same pressure to either plate individually.


                                  edge bonding, is it equivalent to an edge weld? probably not. is it close? from a simulation perspective, i'd say it is. the only way to get closer would be to model the weld. but then you have to start asking yourself about how you're going to model the heat affected zone, the penetration of the weld..etc. it get messy. remember the goal of simulation is to build less prototypes and better understand things at a conceptual level rather than absolute. if you want absolute answers, you can do it, it just takes a lot more time to setup.